Research On Fisheye Camera Calibration And Fisheye Image Matching

In the field of computer vision three-dimensional(3D) reconstruction, ordinary B-Mode ultrasound images have poor gray level, low contrast, and much speckle noise, which lead to missing in significant details and poor results in the3D reconstruction. Traditional3D reconstruction methods based on two-dimensional image can’t meet the requirements of high accuracy, reconstruction speed and the algorithm matching that people’s need in medical and clinical applications. Therefore, in order to solve the problem on the three-dimensional reconstruction of ultrasound scanning data of tissue mimicking material, we carried a more in-depth exploration and research in this paper. Main work of this article is shown below.(1) Present a3D reconstruction algorithm based on an anisotropic diffusion filter. In the Experiment, we get scanned data with ultrasound scanning probe. Firstly, the data should be processed to get71two-dimensional slice images in its order. And then, the images to be filtered and their contours obtained by the canny operator. Finally, the iso-surface rendering method is applied to complete the reconstruction and3D displaying. Subsequently, we present improved method and the crust algorithm as improved algorithm for displaying of3D reconstruction results. Experimental results show the feasibility and accuracy of this method.(2) Present an improved Poisson method for3D reconstruction based on point cloud library. At first, this method converts the data from ultrasound data into point cloud data. Second, to solve the problem of large computation and complex vector fields in Poisson method, we collect the data by uniform interval and eliminate three linear interpolation to simplify the original vector fields. At the same time, eliminating a number of multiple integrals makes computing time greatly improved. Using bilateral filtering before solving the Poisson Equation, the quality of three-dimensional reconstruction can be protected. Experimental results show that the improved method getting a better reconstruction results.Part (1) presents a3D reconstruction algorithm based on an anisotropic diffusion filter and its improvement. The results are better in accuracy and quality of reconstruction. Then, we present an improved crust algorithm for3D displaying. Part (2) presents an improved Poisson method for3D reconstruction based on point cloud library. The accuracy of reconstruction, compute steps and operating time are improved.